Enhancing Password Security With Machine Learning-Based Strength Assessment Techniques

Abstract

This chapter presents a comprehensive evaluation of various machine learning models for password strength assessment. The decision tree, random forest, and AdaBoost models emerge as standout performers, boasting a robust accuracy rate of 84%. Their ability to effectively classify passwords into strength categories demonstrates their value in real-world applications. K-Nearest neighbors, though slightly lower in accuracy, offers a compelling alternative with faster training times and efficient performance. In contrast, Naive Bayes and support vector machine models exhibit limitations, struggling to effectively classify passwords, particularly those of 'medium' strength, despite their speedy training processes. These results underscore the significance of selecting the right machine learning model for password strength assessment, considering factors such as accuracy, training time, and efficiency. In a digital landscape where password security remains paramount, the study's insights provide valuable guidance for enhancing cybersecurity and safeguarding sensitive information.